Adds volume level measurements to the AudioInputController for low-latency clients

media/audio/audio_input_controller.cc

Index: media/audio/audio_input_controller.cc
diff --git a/media/audio/audio_input_controller.cc b/media/audio/audio_input_controller.cc
index 3bdec149bfc3c3605bdb7234edf522199fedb783..c9287449a5db10f54d1fbd56ff93294f9027fa16 100644
--- a/media/audio/audio_input_controller.cc
+++ b/media/audio/audio_input_controller.cc
@@ -5,11 +5,15 @@
 #include "media/audio/audio_input_controller.h"
 
 #include "base/bind.h"
+#include "base/strings/stringprintf.h"
 #include "base/threading/thread_restrictions.h"
+#include "base/time/time.h"
 #include "media/base/limits.h"
 #include "media/base/scoped_histogram_timer.h"
 #include "media/base/user_input_monitor.h"
 
+using base::TimeDelta;
+
 namespace {
 const int kMaxInputChannels = 3;
 
@@ -25,6 +29,20 @@ const int kTimerResetIntervalSeconds = 1;
 // Mac devices and the initial timer interval has therefore been increased
 // from 1 second to 5 seconds.
 const int kTimerInitialIntervalSeconds = 5;
+
+// Time constant for AudioPowerMonitor.
+// The utilized smoothing factor (alpha) in the exponential filter is given
+// by 1-exp(-1/(fs*ts)), where fs is the sample rate in Hz and ts is the time
+// constant given by |kPowerMeasurementTimeConstantMilliseconds|.
+// Example: fs=44100, ts=10e-3 => alpha~0.022420
+//          fs=44100, ts=20e-3 => alpha~0.165903
+// A large smoothing factor corresponds to a faster filter response to input
+// changes since y(n)=alpha*x(n)+(1-alpha)*y(n-1), where x(n) is the input
+// and y(n) is the output.
+const int kPowerMeasurementTimeConstantMilliseconds = 10;
+
+// Time in seconds between two successive measurements of audio power levels.
+const int kPowerMonitorLogIntervalSeconds = 5;
 }
 
 namespace media {
@@ -59,6 +77,7 @@ scoped_refptr<AudioInputController> AudioInputController::Create(
     const std::string& device_id,
     UserInputMonitor* user_input_monitor) {
   DCHECK(audio_manager);
+  DVLOG(1) << "AudioInputController::Create";
 
   if (!params.IsValid() || (params.channels() > kMaxInputChannels))
     return NULL;
@@ -93,6 +112,7 @@ scoped_refptr<AudioInputController> AudioInputController::CreateLowLatency(
     UserInputMonitor* user_input_monitor) {
   DCHECK(audio_manager);
   DCHECK(sync_writer);
+  DVLOG(1) << "AudioInputController::CreateLowLatency";
 
   if (!params.IsValid() || (params.channels() > kMaxInputChannels))
     return NULL;
@@ -173,6 +193,20 @@ void AudioInputController::DoCreate(AudioManager* audio_manager,
                                     const std::string& device_id) {
   DCHECK(task_runner_->BelongsToCurrentThread());
   SCOPED_UMA_HISTOGRAM_TIMER("Media.AudioInputController.CreateTime");
+  DVLOG(1) << "AudioInputController::DoCreate";
+
+#if defined(AUDIO_POWER_MONITORING)
+  // Create the audio (power) level meter given the provided audio parameters.
+  // An AudioBus is also needed to wrap the raw data buffer from the native
+  // layer to match AudioPowerMonitor::Scan().
+  // TODO(henrika): Remove use of extra AudioBus. See http://crbug.com/375155.
+  audio_level_.reset(new media::AudioPowerMonitor(
+      params.sample_rate(),
+      TimeDelta::FromMilliseconds(kPowerMeasurementTimeConstantMilliseconds)));
+  audio_bus_ = AudioBus::Create(params);
+  audio_params_ = params;
+#endif
+
   // TODO(miu): See TODO at top of file.  Until that's resolved, assume all
   // platform audio input requires the |no_data_timer_| be used to auto-detect
   // errors.  In reality, probably only Windows needs to be treated as
@@ -266,6 +300,14 @@ void AudioInputController::DoClose() {
   // Delete the timer on the same thread that created it.
   no_data_timer_.reset();
 
+#if defined(AUDIO_POWER_MONITORING)
+  std::string log_string("AIC::DoClose: closing input stream");

[no longer working on chromium, 2014/05/26 14:20:16]

nit, const

[henrika (OOO until Aug 14), 2014/05/26 15:01:17]

Removed.

+  if (handler_)
+    handler_->OnLog(this, log_string);
+  if (audio_level_)
+    audio_level_->Reset();
+#endif
+
   DoStopCloseAndClearStream();
   SetDataIsActive(false);
 
@@ -377,6 +419,34 @@ void AudioInputController::OnData(AudioInputStream* stream,
   if (SharedMemoryAndSyncSocketMode()) {
     sync_writer_->Write(data, size, volume, key_pressed);
     sync_writer_->UpdateRecordedBytes(hardware_delay_bytes);
+
+#if defined(AUDIO_POWER_MONITORING)
+    {
+      // Only do power-level measurements if an AudioPowerMonitor object has
+      // been created. Done in DoCreate() but not DoCreateForStream(), hence
+      // logging will mainly be done for WebRTC and WebSpeech clients.
+      base::AutoLock auto_lock(lock_);
+      if (!audio_level_)
+        return;
+    }
+
+    // Perform periodic audio (power) level measurements.
+    if ((base::TimeTicks::Now() - last_audio_level_log_time_).InSeconds() >
+        kPowerMonitorLogIntervalSeconds) {
+      // Wrap data into an AudioBus to match AudioPowerMonitor::Scan.
+      // TODO(henrika): remove this section when capture side uses AudioBus.
+      // See http://crbug.com/375155 for details.
+      audio_bus_->FromInterleaved(
+          data, audio_bus_->frames(), audio_params_.bits_per_sample() / 8);
+      audio_level_->Scan(*audio_bus_, audio_bus_->frames());
+
+      task_runner_->PostTask(
+          FROM_HERE, base::Bind(&AudioInputController::DoLogAudioLevel, this));
+
+      last_audio_level_log_time_ = base::TimeTicks::Now();
+    }
+#endif
+
     return;
   }
 
@@ -398,6 +468,24 @@ void AudioInputController::DoOnData(scoped_ptr<uint8[]> data, uint32 size) {
     handler_->OnData(this, data.get(), size);
 }
 
+void AudioInputController::DoLogAudioLevel() {
+  DCHECK(task_runner_->BelongsToCurrentThread());
+
+  // Get current average power level and add it to the log.
+  // Possible range is given by [-inf, 0] dBFS.
+  std::pair<float, bool> result = audio_level_->ReadCurrentPowerAndClip();

[no longer working on chromium, 2014/05/26 14:20:16]

this introduces a race here. audio_level_ is changed/accessed by both the
real-time audio thread and task_runner_ thread.

Instead, you should do all the calculation in the real-time audio thread and
just and post the values to task_runner_ thread.

[henrika (OOO until Aug 14), 2014/05/26 15:01:17]

Will fix.

+  std::string log_string = base::StringPrintf(
+      "AIC::OnData: average audio level=%.2f dBFS", result.first);
+  static const float kSilenceThresholdDBFS = -72.24719896f;
+  if (result.first < kSilenceThresholdDBFS)
+    log_string += " <=> no audio input!";
+  if (handler_)
+    handler_->OnLog(this, log_string);
+
+  // Reset the average power level (since we don't log continuously).
+  audio_level_->Reset();
+}
+
 void AudioInputController::OnError(AudioInputStream* stream) {
   // Handle error on the audio-manager thread.
   task_runner_->PostTask(FROM_HERE, base::Bind(